You are on page 1of 6

Journal of ELECTRONIC MATERIALS, Vol. 44, No.

6, 2015

DOI: 10.1007/s11664-014-3560-0
 2014 The Minerals, Metals & Materials Society

High-Temperature Thermoelectric Properties of (1  x)


SrTiO3  (x) La1/3NbO3 Ceramic Solid Solution
DEEPANSHU SRIVASTAVA,1,3 F. AZOUGH,1 M. MOLINARI,2
S.C. PARKER,2 and R. FREER1
1.School of Materials, University of Manchester, Manchester M13 9PL, UK. 2.Department
of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, UK. 3.e-mail: deepanshu.
srivastava@manchester.ac.uk

Ceramics based on SrTiO3 are of growing interest as thermoelectric materials


because of their high-temperature stability and non-toxicity. Substitution of
La and Nb into the perovskite structure provides opportunities to control both
the microstructure and properties. Ceramic solid solutions of (1  x)
SrTiO3  (x) La1/3NbO3 were prepared by the mixed oxide route, using compositional steps of x = 0.1. Pressed pellets were sintered at temperatures of
1573 K to 1723 K in air. Addition of aliovalent ions (La3+, Nb5+) on the A/B
sites (Sr2+, Ti4+) led to A-Site cation deficiency in the stoichiometric compositions and other defect structures which increased carrier concentration. A
maximum ZT of 0.004 was obtained for the x = 0.2 stoichiometric sample,
although much higher ZT values are possible by sample reduction.
Key words: SrTiO3, La1/3NbO3, thermoelectric, cation deficiency, figure of
merit

INTRODUCTION
With the rapid depletion of fossil fuels and the
increasing demand for energy, renewable sources of
energy and ways to increase energy efficiency are of
growing importance. For over 50 years, thermoelectric generators (TEGs) have been employed
which are based upon the Seebeck effect,1 whereby
a potential difference is developed across a material
when its ends are held at different temperatures.
TEGs can be used to generate electrical power from
environments where there are temperature gradients,2 and can be utilised to convert waste heat
energy into usable electrical energy. The conversion
efficiency (g) is dependent on the temperature difference (between Tcold and Thot) and a dimensionless
term, ZT, defined as the figure of merit for the
material, given by Eqs. 1 and 2 below3:
p

1 ZT  1
 (1)
g conversion g carnot  p
1 ZT TTcold
hot
efficiency
efficiency
(Received July 15, 2014; accepted November 24, 2014;
published online December 18, 2014)

ZT

S2 r
T
j

(2)

where S is the Seebeck coefficient, r is electrical


conductivity, j is thermal conductivity of the sample, and T is the temperature.
Traditional thermoelectric materials are frequently metallic, and include Pb-based tellurides,4,5
and CoSb3-based skutterudites.6,7 Whilst these
materials have been effective under specific environments, usually low temperature, their susceptibility to degradation at high temperature has
restricted their applications. In contrast, oxide
materials are stable and offer distinct stability
advantages over conventional bulk materials at
higher temperatures. Furthermore, oxides are generally cheaper, more environmentally safe and
abundant compared to rare elements used in traditional thermoelectric materials.8,9 The interest in
oxides was spurred after the discovery of impressive
thermoelectric properties in layered cobaltites,
which are often p-type materials.10,11 Over the past
decade, SrTiO3-based materials have emerged as
one the most promising n-type materials with the
highest ZT value of 0.37 following Nb doping.12
1803

1804

Srivastava, Azough, Molinari, Parker, and Freer

Doping of SrTiO3 is critical to maximise thermoelectric performance.1315 In this investigation, we


focus on the solid solution of SrTiO3 and La1/3NbO3.
SrTiO3 is known to have a stable perovskite cubic
Pm3m structure,16 while La1/3NbO3 has an orthorhombic Cmmm structure at 300 K.17 As La enters
the perovskite A-site, and Nb enters the perovskite
B-site, the addition of La1/3NbO3 to SrTiO3 simultaneously enables site modification and generation
of vacancies on both sites. The thermoelectric
properties of solid solution: Sr1xLax/3Ti1xNbxO3
was investigated for values ranging between 0 and
0.3.
MATERIALS AND METHODS
The starting powders used were oxides and carbonates of Sr, Ti, La, and Nb. The powders were
obtained from Sigma Aldrich (UK) (SrCO3), PowerWave Technologies (UK) (La2O3, TiO2), and Solvay
(UK) (Nb2O5); all having >99.9% purity. Prior to
use, the La2O3 was heat-treated at 1173 K for 10 h
to eliminate CO2 and moisture. The powders were
mixed in appropriate molar ratios to produce the
Sr1xLax/3Ti1xNbxO3 solid solutions. The samples
prepared include SrTiO3 (denoted as STO), and
Sr0.8La0.067Ti0.8Nb0.2O3,
Sr0.9La0.034Ti0.9Nb0.1O3,
and Sr0.7La0.1Ti0.7Nb0.3O3 denoted as L1, L2, L3,
respectively. An excess of 0.5 wt.% of TiO2 was
added to the starting powders as sintering aid.
Propan-2-ol and yttria-stabilized zirconia milling
balls were added in equal weights to the powder mix
and vibration milled for 24 h. The slurry was dried
and calcined at 1373 K, then milled again for 24 h to
reduce average particle size. The dried powders
were pressed uniaxially in a hardened steel die of
20 mm diameter to produce cylindrical pellets 5 mm
thick. The pellets were sintered in a Vecstar box
furnace (model No. XF1) under normal pressure and
oxidising conditions at 1700 K for 8 h. The densities
of the sintered ceramics were determined using the
Archimedes method. Circular cross-sections of
individual specimens were cut using a diamond
wheel, then ground and polished down to a 1-lm
diamond paste followed by silica solution (OPS)
ready for phase and microstructural analysis. X-ray
diffraction (XRD) was carried out using a Philips
Xpert diffractometer with Cu source. The pattern
was recorded continuously by scanning 2h angle
from 10.0 to 85.0 with steps of size 0.030 degrees
and dwell time of 10.00 s at each step. Scanning
electron microscopy (SEM) was undertaken on a
Phillips XL30 FEGSEM. The chemical analysis on
the microstructure at 20 keV was carried out by a
Rontec EDX (energy dispersive spectroscopy) system available in conjunction with XL30 FEGSEM.
The SEM micrographs obtained were used to
determine the average grain size by the linear
intercept method.18 The electrical conductivity and
Seebeck coefficient were determined from 550 K to
1050 K using a ZEM-III (manufactured by Ulvac-

Riko). The experimental setup to determine thermal


diffusivity was developed in-house using a laser
flash method.19 The Cowan method20 was used to
model the compensation in heat loss and obtain
temperature-dependant thermal diffusivity values
of all samples in a temperature range from 300 K to
1100 K. Specific heat capacity was determined by
differential scanning calorimetry (Netzsch STA 449
C) where specific heat flow into a material is measured with respect to known reference material to
obtain the heat capacity of the material. For transmission electron microscopy, samples were prepared
by a crushing method. The sintered disks were
crushed to powder using an agate mortar and pestle. Grains of individual powders were dispersed in
chloroform, dropped onto a copper grid with holey
carbon film, and then dried. Atomic level resolution
structural characterization was carried out using an
aberration-corrected FEI TEM-STEM (80200 keV
ChemiSTEM) operating at 200 keV. High-angle
annular dark field (HAADF) imaging was performed using an annular detector with an inner
angle of 100 mrad and an effective outer angle of
185 mrad.
RESULTS AND DISCUSSION
The densities for all samples sintered in air were
above 95% of the theoretical value. XRD analysis of
the sintered Sr1xLax/3Ti1xNbxO3 solid solutions
revealed that specimens having formulations based
on x > 0.3 did not stabilise as predominantly single
phase products and could not be sintered at temperatures less than 1723 K. The investigation
therefore focussed on sample formulations with
0 x 0.3. XRD spectra for the relevant specimens
(STO, L1, L2 and L3) are presented in Fig. 1. All
composition were indexed as a cubic perovskite
structure belonging to space group Pm3m. The lat for
tice parameter increased linearly from 3.905 A
for L1, to 3.9151 A
for L2, and to
STO, to 3.9114 A
for L3 composition upon La and Nb doping.
3.919 A
The variation in lattice parameter is consistent with
the Vegards Law,21 and the introduction of dopants
of larger ionic radii;22 lattice parameter data are
shown as the inset of Fig. 1.
Typical examples of microstructures are shown as
backscattered SEM micrographs in Fig. 2. The base
material, STO (Fig. 2a), exhibits average grain size
of 5.1 lm. For the doped samples L1, L2, L3, the
increase in grain size is evident (Fig. 2bd) with
average grain sizes of 9.9 lm, 12.7 lm, and 21.1 lm,
respectively. Grain growth (in L1, L2 and L3) for the
same sintering conditions as for STO is also indicative of a lowering of the effective sintering temperature by doping. Another interesting feature is
the optical contrast, with bright spots at the
centre of few grains of L1 and L2 compositions
(Fig. 2b and c). The spots are smaller than the
minimum interaction volume for energy dispersive
spectroscopy (EDS) to give convincing results.

High Temperature Thermoelectric Properties of (1  x) SrTiO3  (x) La1/3NbO3 Ceramic Solid Solution

1805

Fig. 1. X-ray diffraction spectra for compositions in the Sr1xLax/3Ti1xNbxO3 solid solution.

Fig. 2. Backscattered electron SEM images of samples sintered in air: (a) STO, (b) L1, (c) L2, (d) L3. Note lower magnification used in (d), to
image the grain distribution and second phase.

However, the bright Z contrast suggests that there


is segregation, with a composition of higher atomic
weight at the core. Such segregation may reflect
the remnants of an intermediate phase formed
during calcination. However, in terms of thermoelectric performance, such coreshell structuring
and presence of additional interfaces may be beneficial if they can reduce thermal conductivity. In
contrast, the L3 samples as shown in Fig. 2d did
not exhibit distinct cores, but a higher atomic
weight secondary phase was observed along the
grain boundaries. EDX analysis confirmed the
grain boundary secondary phase was rich in Nb
and La.
High-resolution transmission electron microscopy
was used to investigate the distribution of A-site

and B-site atoms and vacancies in samples of


Sr0.7La0.1Ti0.7Nb0.3O3 (L3). Figure 3a shows a
h1 0 0i orientation HAADF images for L3. In this
orientation, we are effectively looking down columns
of atoms, and the brightness of the spot denoting a
column indicates the average atomic number of
atoms in the columns; brighter means higher average atomic number5 in the column. The very faint
column shown in a circle near the top of the image
indicates a vacancy-rich column. The presence of
rows of A-site and B-site ions are identified near the
centre of the image (Fig. 3a). Detailed intensity scans
for these rows are shown in Fig. 3b and c. The uneven variation suggests that there is no evidence of
ordering of La or Nb on this scale. However, there is
evidence of vacancy-rich columns (such as the one

1806

Srivastava, Azough, Molinari, Parker, and Freer

Fig. 3. (a) HAADF (Zone axis: h1 0 0i) image for L3 showing both A-site and B-site atomic columns. The variation in the optical intensity along
the rows indicates random distribution of Sr, La and vacancies in A-site and Ti and Nb in B-sites; (b) detailed intensity scan along A-sites;
(c) detailed intensity scan along B-sites.

indicated in Fig. 3a), and such defect structures


reflect local variations in mass which can hinder
phonon propagation at higher temperatures and help
to reduce thermal conductivity.
The thermoelectric properties of the samples
determined at temperatures in the range 550
1050 K, are presented in Fig. 4.
There was wide variation in the resistivity of the
samples (Fig. 4a). The electrical conductivity
(reciprocal of resistivity) of the STO followed semiconductor behaviour reflecting the wide band gap of
3.2 eV.23 Due to the high resistivity, data for the
STO sample could not be reliably collected below
750 K. As La and Nb were substituted into the A
and B sites of SrTiO3, respectively, the resistivity of
the doped samples at 1000 K reduced from 30 ohmm (for STO) to 1 9 101 ohm-m for L1 and to
5 9 102 ohm-m for the L2 and L3 samples. Such
changes in resistivity could be attributed to electronic structure modulation caused by La and Nb
doping, as the A-site cation deficiency created by
doping will affect the Ti4+/Ti3+ ratio,23 and will be
compensated by the creation of oxygen vacancies.
This mechanism would explain the lower electrical

resistivity of L1/L2/L3 compared to bulk STO.


Measurements of the concentration of charge carriers are needed to verify the conduction mechanism
operating in the samples. Even though there is no
clear trend in the reduction in resistivity for samples in (1  x) SrTiO3  (x) La1/3NbO3 system for
x > 0, it is inferred that a higher A-site deficiency
will increase charge carrier concentrations. It
should be noted that the increase in grain size with
increase in x would marginally improve the conductivity by reducing grain boundary resistance.
Indeed, the grain boundary second phases in L3
samples could have increased the resistivity against
the above trend.
Undoped STO exhibits a very high Seebeck coefficient which is reduced when alloyed with La1/3
NbO3; for example, it reduced from 1050 lV/K to
800 lV/K for L1 formulation at 1000 K (Fig. 4b).
Heikes Law24 indicates that the Seebeck coefficient
is inversely proportional to the electronic charge
carrier concentration. The Seebeck values for all
three doped samples were negative, indicating
electrons as the predominant charge carriers. Furthermore, the Seebeck coefficient did not vary

High Temperature Thermoelectric Properties of (1  x) SrTiO3  (x) La1/3NbO3 Ceramic Solid Solution

1807

Fig. 4. Thermoelectric properties of samples in the (1  x) SrTiO3  (x) (La1/3NbO3) system sintered in air. Temperature dependence of (1  x)
SrTiO3  (x) (La1/3NbO3) is shown as (a) resistivity, (b) Seebeck coefficient, (c) thermal conductivity and (d) figure of merit, ZT. Compositions
STO, L1, L2, L3 corresponds to x = 0.0, 0.1, 0.2, 0.3, respectively.

significantly with temperature; the values followed


the trends resistivity data for all the samples.
From the measured electrical properties, uniform
power factors, typically 5.5 9 106, 1.1 9 105 and
7.5 9 106, were deduced for L1, L2, L3 with
respect to temperature. The samples reported here
have not been doped with excess charge carriers
(essentially stoichiometric compositions) and were
sintered in oxidizing environments. Much work
can be done to improve the electrical transport
properties by controlled excess doping of La/Nb. The
control of oxygen non-stoichiometric can also be
useful in obtaining very high power factors, similar
to those achieved in heavily doped oxides.25
The main advantage of the (1  x) SrTiO3  (x)
La1/3NbO3 formulations is evident from the reduction of thermal conductivity (Fig. 4c). For bulk STO
materials, the thermal conductivity is intrinsically
high and detrimental to the thermoelectric applications. Here, the thermal conductivity of the STO
samples at 600 K to 1000 K was between 6.5 W/mK
and 4 W/mK, consistent with published data for
polycrystalline STO samples.26 The incremental
addition of La1/3NbO3 had a positive effect, leading
to a reduction in thermal conductivity values. This
was attributed to A-site vacancies and the presence
of the larger La and Nb ions in the crystal structure.
These structural modifications act as scattering

centres for phonon propagation at high temperatures. Moreover, the clustering of vacancies and the
uneven distribution of vacancies or La/Nb ions
(Fig. 3a) creates disorder in the crystal structure,
assisting phonon scattering across a wider spectrum. The development of second phase segregation
in sample L3 could also make a significant contribution to reducing lattice thermal conductivity,
yielding the lowest thermal conductivity of 2.5 W/
mK in sample L3 at 1000 K. The mean free path for
phonon conduction in STO-based compositions is
reported to be in the range of 0.210 nm which
accounts for most of the heat conduction in bulk
samples.26,27 It should be possible to achieve further
reduction in the overall thermal conductivity by
nanostructuring the grain structure and by developing nanoinclusions at grain boundaries or nanoprecipitates within the grains.
From the thermoelectric parameters determined
at 5501050 K, the ZT values were determined; the
highest ZT value of 0.004 was obtained for L2
sample as seen in Fig. 4d. The value is very low due
to fact that no excess charge carrier had been added.
However, it defines the best composition in studied
solid solution to investigate further by controlled
nanostructuring and generating excess charge carriers within the material. For example, a pilot study
of L2 sintered in reducing conditions gave a ZT of

1808

Srivastava, Azough, Molinari, Parker, and Freer

0.29. This investigation forms the next phase of the


programme of work.
CONCLUSIONS
High density samples in the solid solution (1  x)
SrTiO3  (x) La1/3NbO3 system for 0 x 0.3 were
obtained by mixed oxide processing. The crystal
structures were cubic and the lattice spacing
increased with change in composition. Without excess
doping of charge carriers from primary components,
the La/Nb-doped L1, L2 and L3 samples exhibited
improved conductivity by at least an order of 2 at
1000 K, without much reduction of Seebeck coefficient. The thermal conductivity of samples with
0 < x 0.3 was lowered by at least 25% at 1000 K for
L1 sample, while the L3 sample showed lowest thermal conductivity of 2.5 W/mK at 1000 K. These
improvements in properties are attributed to structural modifications, and A-site vacancies. Based on
microstructural characteristics and thermoelectric
properties, the L2 samples were identified as optimum
amongst the samples examined. Controlled doping
and tuning of oxygen vacancies improved the power
factor compared to bulk STO. Use of excess doping to
generate additional charge carriers is essential to
generate ZT > 0.3. The effect of nanoinclusions and
second phases in further reducing thermal conductivity will be investigated in future studies.
ACKNOWLEDGEMENTS
D.S. acknowledges support of a Presidents Scholarship from the University of Manchester. The
SuperSTEM facilities are supported by the EPSRC.

REFERENCES
1. T.J. Seebeck, Ueber die magnetische Polarisation der Metalle und Erze durch Temperatur-Differnz. Berlin: gedruckt
in der Druckerei der Konigl. Akademie der Wissenschaften.
26 cm (1825).
2. T.M. Tritt, Annu. Rev. Mater. Res. 41, 433 (2011).
3. D.M. Rowe and C. Chemical Rubber, CRC Handbook of
Thermoelectrics [electronic resource]. Boca Raton, FL: CRC
Press, 1995. 1 online resource (701).
4. G.S. Nolas, in Materials Research Society Symposium Proceedings, November 2530, 2012, Boston, MA, ed. by
Y. Grin, D.C. Johnson, and A.J. Thompson (Warrendale:
Materials Research Society, 2013), pp. xiv, 245.
5. Y. Pei, et al., Energy Environ. Sci. 4, 2085 (2011).
6. T. Caillat, J.P. Fleurial, and A. Borshchevsky, J. Phys.
Chem. Solids 58, 1119 (1997).
7. G.S. Nolas, et al., J. Appl. Phys. 79, 4002 (1996).
8. J.W. Fergus, J. Eur. Ceram. Soc. 32, 525 (2012).
9. K. Koumoto, et al., J. Am. Ceram. Soc. 96, 1 (2013).
10. I. Terasaki, Phys. B 328, 63 (2003).
11. K. Koumoto, I. Terasaki, and R. Funahashi, MRS Bull. 31,
206 (2006).
12. K. Koumoto, et al., Annu. Rev. Mater. Res. 40, 363 (2010).
13. T. Okuda et al., Phys. Rev. B 63 (2001).
14. S. Ohta, et al., J. Appl. Phys. 97, 034106 (2005).
15. N. Wang, et al., J. Ceram. Soc. Jpn. 118, 1098 (2010).
16. O. Madelung, U. Rossler, and M. Schulz (ed.), SrTiO3 crystal
structure, lattice parameters.
17. A.N. Salak, et al., Appl. Phys. Lett. 93, 162903 (2008).
18. M.I. Mendelson, J. Am. Ceram. Soc. 52, 443 (1969).
19. W.J. Parker, et al., J. Appl. Phys. 32, 1679 (1961).
20. R.D. Cowan, J. Appl. Phys. 34, 926 (1963).
21. A.R. Denton and N.W. Ashcroft, Phys. Rev. A 43, 3161 (1991).
22. R.D. Shannon, Acta Crystallogr. Sect. A 32, 751 (1976).
23. W. Wunderlich, H. Ohta, and K. Koumoto, Phys. B 404, 2202
(2009).
24. R.R. Heikes and R.W. Ure, Thermoelectricity: Science and
Engineering (New York: Interscience Publishers, 1961),
p. 576.
25. C. Yu, et al., Appl. Phys. Lett. 92, 092118 (2008).
26. W. Yifeng, et al., Appl. Phys. Express 3, 031101 (2010).
27. G.G. Yadav, et al., Nanoscale 3, 4078 (2011).